This study investigates communicative performance in people with severe restriction in verbal output due to non- fluent aphasia. Such individuals are trained to use an alternative communication program called C-Speak Aphasia. This program is picture-based and operated on a Mackintosh computer. Using C-Speak, patients learn to select icons from semantic-category groups and combine them to form novel messages that can be spoken aloud by the computer s speech synthesizer. This project proposes to conduct a controlled study of treatment outcomes in people with aphasia who are trained to use C-Speak. The study uses a multiple baseline design to determine whether patients improve communication on several tasks while using C-Speak, compared to communication without C-Speak. Communication performance is probed repeatedly during a 4-6 month training period with 5 tasks; providing autobiographical information, describing pictures and observed events, communicating via writing, and telephone communication. Linguistic factors related to semantic processing and cognitive factors related to executive functioning are hypothesized to be related to individual aphasic patients abilities to communicate using C-Speak. These factors will be assessed during baseline testing, and used as predictors in a multiple regression analysis to investigate whether they predict changes in participants ability to use C- Speak for functional communication. Finally, two experiments relating to semantic functioning and executive system function will also be completed using the subjects who undergo C-Speak training. Scores on these measures will also be used as predictors of successful C-Speak usage.